Tuesday 18 December 2007

PhD Matters

While talking with another PhD candidate (as you do) about my proposed work in genetic algorithms and evolutionary programming, I gave a guess about how my method would be likely to converge. That is, how long it would take for evolution to get a "good answer" assuming all prejudices we had about how to make it get better, quicker were removed.

I thought it would take a long time frommaging around, not really improving a lot, until a truly advanced mutation happened, and then it would take off, at least until the definition of "good result" changed.

Then it struck me. If my intuition is right - and it may not be, only the experiments will show that - I think I'm predicting that punctuated equilibrium is not just possible, but inevitable in this kind of evolutionary programming.

Now that doesn't mean that biological evolution necessarily follows that model. But as I'm attempting to make a general technique, inefficiently applicable to every kind of problem, it's an exciting insight. That if we remove all a priori knowledge, allowing full scope for "emergent behaviour", surprising and unintended solutions, then it's an inevitable consequence.

More to the point, my intuitive conjecture can be tested. We can get hard data to support or refute it. Either is good.

For the first time for ages, I think I'll really be enjoying this research, and will be good at it too. The unfortunate circumstances that have forced me to re-start from scratch could be a blessing in disguise. A very effective disguise, true.

I better explain. I was (past tense) involved in the "software complexity" project, under the auspices of the Cooperative Research Centre for Advanced Automotive Technology (AutoCRC). Research directed by, and to the benefit of, a particular industrial partner. Then that partner disappeared from the scene, leaving us high and dry, our research of no interest to other CRC members. So it had to be abandoned, and the PhD students working on it had to come up with another research area, one that excited the interest of the AutoCRC, get it approved, and re-start from scratch.

I put in my proposal two weeks ago, over a video conference and accompanied with much hand-waving, and got it accepted immediately. I may not be as bright as some of the other PhD candidates I'm working with, but I give Good Presentation. :)

And during this period, I was parenting a 6 year old boy, travelling to Thailand and recovering from major surgery, recovering from the necessary operation to revise that surgery, battling the bureaucracy to get a passport, blogging, taking a side-trip to Israel to do some work at Haifa Naval Base, and of course still learning how to be me.

It's been said that those women and men who transition, while they have more problems than before, become far more capable of dealing with them. That they blossom, in personality and capability. Now I hadn't noticed that, but looking at the list of what I've managed to do, maybe it's true. Gosh!

2 comments:

mythusmage said...

That you now see things in a new way is likely help. As well as the fact you read expressions and body language better than you had before. You ladies do have talents we guys don't.

Good luck with the PHD work, and tell the puppy Mythusmage said, "Hi!"

Anonymous said...

Congratuations!